This article will help you:
Amplitude's Event Segmentation chart helps you understand what specific groups of users are doing in your product. For example, in an event segmentation analysis, you can:
- Identify the top events fired over a selected time period
- Compare event totals to each other
- See which users fire certain events
Analyzing your event segmentation data takes place in the Metrics Module, the bottom panel of your screen. This is where your events and user segments come together to give you a visualization of user behavior.
Before you begin
If you haven't done so already, you'll want to familiarize yourself with the basics of building charts in Amplitude , as well as how to create an Event Segmentation chart. You can also see FAQs about the Event Segmentation chart here.
Interpret your Event Segmentation chart
Event Segmentation is Amplitude's most commonly-used chart. It's simple enough to create a useful analysis quickly, even as a brand-new Amplitude user. But it also has many powerful features that will add depth and nuance to your analysis. The rest of this article will explore these features and explain how you can best put them to use to generate insights on user behavior.
Choose the right metric
Amplitude offers you several different ways of looking at your event segmentation results. In this section, we'll explain the differences between them.
The default metric for the Event Segmentation chart, it displays the total count of unique users in your segment who fired the event you added in the Event Module. View the exact count by simply hovering over the specific data point you’re interested in. If you want to inspect the users who make up that data point, just click on it (see our Help Center article on Amplitude’s Microscope feature to learn more).
Like Uniques, Event Totals is a straightforward, count-based metric. The difference is that instead of counting unique users, it graphs the total count of times a specific event was fired at each data point.
This metric graphs the percentage of all active users (defined as users who have fired any active event in a specified time frame) who fired a specific event at each data point.
The Average metric graphs the average number of times a specific event was fired. Here, the "average" for any data point is equal to its event totals divided by unique users.
When you apply the Frequency metric, Amplitude will group the users included in your user segment into buckets defined by the number of times each has fired an event during the time frame of your analysis.
Here, we see an event segmentation analysis using the Frequency metric. Each stacked area represents a "frequency bucket." For each data point, Amplitude displays the number of users contained in that bucket. And as described above, if you want to learn more about the users in a particular data point, all you have to do is click on it.
In the screenshot above, the default buckets are represented by the colored dots. Click customize buckets to adjust the sizing of the buckets and distribution of the data, or use the Custom Buckets modal to set individual ranges for each bucket.
Depending on the details of your analysis, you may also be able to generate an event segmentation chart based on the values of your event or user properties.
- Sum of Property Value: Graphs the sum of property values at each data point. To use this metric, the property value must be an integer.
- Distribution of Property Value: Shows the distribution of event totals broken out by the values of the selected event property. The minimum value is inclusive, and the maximum value is exclusive.
- Average of Property Value: Graphs the average of the property values, or the sum of those values divided by the total number of events fired at each data point. To use this metric, the property value must be an integer.
- Distinct Property Values per User: Graphs the average count of different property values triggered by each user. More specifically, it's the total sum of unique user-distinct property value pairs, divided by the number of users.
- Median Property Value: Graphs the median property values for each data point. This is most useful in situations where averages might be noticeably skewed by outliers. To use this metric, the property value must be an integer.
In an Event Segmentation chart, you can write formulas that Amplitude will apply to the events you've included in your analysis. To read more about each formula and see some examples of use cases, see our Custom Formulas article.
Below the chart, you'll see a table of the data included in your chart. Within the data table, you can:
- Update the chart display, by selecting or deselecting rows in the table
- Download the data table, by clicking the "Export CSV" button
- Modify the summary column based on Row Average, Median, Change, or Sum(Only available for Event Totals, Properties, and Formulas)
- Sort by columns, by clicking the column name (i.e. Sum, May 21)
By default, Amplitude will include all top values or events in this table, which will update automatically when Amplitude receives new top values or events. You can turn this off by first deselecting the segments and then explicitly selecting the values and events you want to keep.
Change your chart view
Whichever metric you choose, you’ll have several options when it comes to how you want your results displayed on the chart.
The default setting is a basic line chart. These are useful for looking at the trend of one event for one user category over time.
Stacked area charts are useful when you’re looking at data that breaks down into discrete buckets, like when you’re analyzing multiple events.
Bar charts are good for situations when you want to show a distribution of data points, or compare metric values across different segments of your data. Bar charts make it easy to see which values are highest or most common, and how specific groups compare against the rest.
A stacked bar chart will show how broad categories or buckets are divided into smaller ones, as well as the relationship each of those smaller parts has to the overall total.
If your analysis uses multiple "group by" conditions, the resulting visualization might turn out confusing and hard to interpret. While it’s not specifically a chart type, the group by visualization will clarify your data in these circumstances.
For example, here the control panel groups the 'Play Song or Video' event by genre type, as well as by country and platform.
Instead of a temporal or “bucket-based” chart, this visualization generates a table view, breaking out the genre type, country and platform (in other words, the “group by” conditions that were applied to this analysis) into separate columns. This makes it easier to digest and cross-reference the data. In this example, most users in the United States who played pop songs were doing so from a web platform.
For more information on the syntax and limitations of group by conditions, see our Help Center article here.
Switch between absolute totals and relative percentages
When using stacked area charts and stacked bar charts, you can choose to view your analysis in terms of relative percentages instead of absolute totals.
# Absolute will display the overall user volume, whereas % Relative gives you the series value divided by the sum of all the series values.
Advanced features: averages, windows, and cumulative totals
Rolling averages will display the unweighted mean, which works to "smooth out" a chart. This is useful if you have cyclical users—for example, people who use your product during the week, but not on weekends.
To apply a rolling average to your chart, click Advanced and select Rolling Average from the drop-down list.
This chart displays the daily event totals between February 5th and March 7th, without a rolling average. Note the sharp peaks and valleys in the line.
But when a rolling average of seven days is added, those fluctuations disappear. That’s because each data point is now an average of the previous seven days’ worth of data.
Bear in mind that each day’s data is included in that day’s data point. When looking at the current day, Amplitude will use a dotted line to show data collection for today isn’t finished yet. You can hide the dotted line by using Offset in the date picker.
Note also that with a seven-day rolling average, the first six days of your selected time frame will fetch data from outside the selected time period. For example, in an analysis covering the month of February, the result for February 6th would average data over January 31st to February 6th.
A rolling window is another method of “smoothing out” your data. It will display the aggregate last N days of information in a single data point. This is useful if you want to see aggregated metrics—such as your 7-day active user count—on a daily basis.
This differs from the rolling average, in that a rolling window does not average your data over the selected time frame. Instead, it sums the data.
To apply a rolling window to your chart, click Advanced and select Rolling Window from the drop-down list.
This chart displays daily uniques between April 5th and May 5th without a rolling window. With Microscope, we can see that on April 21st, there were 173,144 users.
Below, we see the daily uniques displayed with a rolling window of seven days. The April 21st data point is the number of unique deduplicated users between April 15th and April 21st.
As with a rolling average, when using a seven-day rolling window, the first six days of your selected time frame will fetch data from outside the selected time period. For example, in an analysis covering the month of February, the result for February 6th would average data over January 31st to February 6th.
Cumulative sum will display a running total of events in a single data point. For example, you might want to show a running total of revenue generated by purchase events. Cumulative sum will help you do that.
To apply a cumulative sum to your chart, click Advanced and select Cumulative from the drop-down list.
NOTE: If you would like to use cumulative sum in a formula, click Formula and type out CUMSUM.
This chart shows a running total of revenue generated by "Complete Purchase" events. The April 19th data point represents a sum of revenue generated on all the preceding days of the selected time frame. Here, that means April 17th, April 18th, and April 19th.
Using cumulative sum with uniques will generate a deduplicated count of unique users for each data point.
- On 4/17, User A fired "Complete Purchase".
- On 4/18, User A and User B fired "Complete Purchase".
- On 4/19, User C and User D fired"Complete Purchase".
On the data point for 4/19, a total count of four will be returned because four unique users fired this event from 4/17 to 4/19.
You can view segmentation data in real time. However, there are some caveats:
- You can only segment one day's worth of data for real-time
- The event times are rounded down
- Charts are cached every five minutes for everyone
Period-over-period comparison (compare to past)
Using the Compare to past feature, you can compare the results of the current time range with the previous day, or the same day from previous week, month, quarter or year.
For example, let's say you want to compare the daily active users for the current week to last week.
The blue segment shows you the current period and the green segment shows you your data for last week.
Since the period-over-period comparison interval is configurable, you can choose what dates you actually want to compare. You can also toggle to see the percentage change between values instead.
Period-over-period for custom formulas
You can also use the period-over-period comparison with the custom formula metric. For example, you can compare your current rolling average with that of the previous month:
You can also see FAQs about the Event Segmentation chart here.